Skip to content
Arjit Kulkarni
01Production Backend

Credit Claim Service

Replacing a manual credit-grant workflow at production scale
Role
SDE Intern — Reskilll (Practern Technology Pvt. Ltd.)
Context
May 2026 — Present · Bengaluru, India
Stack
PythonFastAPIPostgreSQLGitCI/CD
Source
Private
01  The problem

A FastAPI service that turned a hand-run credit-grant process into an auditable, self-serve flow for a developer portal with 500+ users.

Credits on a 500+ developer portal were granted by hand. The process was slow, easy to get wrong, and produced duplicate grants that only surfaced during a recurring manual audit — which meant reconciling state that lived across several services with no single source of truth.
02  Approach
  1. 01

    Designed REST APIs and PostgreSQL data flows covering the full lifecycle of a claim: submission, validation, grant, and reconciliation.

  2. 02

    Traced state across services to root-cause the duplicate credit grants, rather than patching the symptom at the edge.

  3. 03

    Wrote Python reconciliation scripts that validate records end to end, turning the recurring manual audit into an automated check.

  4. 04

    Profiled the PostgreSQL query plans behind the hot REST endpoints, then applied indexing and query restructuring where the profile pointed.

  5. 05

    Shipped incrementally through Git-based development, code review and CI/CD, working directly with product and operations stakeholders.

03  Outcome

  • A manual credit-grant workflow replaced by a shipped, production service.

  • Duplicate grants root-caused and eliminated, removing a recurring manual audit step.

  • Average query latency reduced by 30% across production REST endpoints.

Developers on the portal
0+Developers on the portal
Lower avg. query latency
0%Lower avg. query latency
Manual audit steps left
0Manual audit steps left

Concepts applied

  • API design
  • Data modelling
  • Query optimisation
  • Root-cause analysis
  • Incremental delivery

Next project

02LSM-KV
All work